Hello,
I have a question concerning the input FA images for TBSS.
Our group collects an axial DTI from a Siemens Trio machine. We convert the
data to Analyze format using mri_convert and use FDT for DTI processing.
After mri_convert, the data remain in radiological orientation (avworient
-gerorient).
When I run TBSS on this data, there are major alignment problems
(over-warping), as seen in the stats directory with mean_fa and all_fa.
However, if the original DTI data is first aligned to the MNI brain (6dof
affine flirt) and then used as the input for TBSS, I no longer have
problems with TBSS registration (ie all_fa and mean_fa look fine).
My assumption is that the final areg, target to MNI transformation in
tbss_3_postreg does not use a -180, 180 search, and thus is unable to
properly register the data to the MNI brain. Is this correct?
Any recommendations for input FA images (eg, use the 6 parameter flirt as
described above, use avwswapdim to properly orient the data to the MNI
brain, etc).
Thanks in advance for any help!
-Ryan
|